Without blaming Hetzner in any way for this, this is the reason why we pay the GCP premium. We hosted at various providers in the past and there have always been problems. At one time it was because the A/Cs weren't working. And that was a big provider.
At GCP we never had any problems with any outage so far. Even with VMs or the hosted features such as Cloud SQL. Whenever there is a maintenance, the systems keep working seamlessly (they migrate the systems somewhere else whenever they do maintenance work or something).
We never felt the need to create high availability services, because it was never necessary so far (the services have always been available). And we've been hosting there for years.